CN114900246B - Noise substrate estimation method, device, equipment and storage medium - Google Patents
Noise substrate estimation method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN114900246B CN114900246B CN202210572981.1A CN202210572981A CN114900246B CN 114900246 B CN114900246 B CN 114900246B CN 202210572981 A CN202210572981 A CN 202210572981A CN 114900246 B CN114900246 B CN 114900246B
- Authority
- CN
- China
- Prior art keywords
- iterative
- stage
- iteration
- filtering
- threshold
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B15/00—Suppression or limitation of noise or interference
- H04B15/005—Reducing noise, e.g. humm, from the supply
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/309—Measuring or estimating channel quality parameters
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Abstract
The invention discloses a noise substrate estimation method, a device, equipment and a storage medium, wherein the method comprises the following steps: calculating a signal power spectrum in response to the acquired signal; performing first-stage iterative filtering through a corrosion filter; detecting spectral entropy white noise frequency points of signals; calculating a data updating threshold and an iteration convergence threshold of the first stage of iterative filtering; repeatedly executing the first stage iterative filtering, detecting spectral entropy white noise frequency points of signals, calculating a data updating threshold and an iterative convergence threshold of the first stage iterative filtering until corrosion filtering iteration is completed, and recording total iteration times and noise floor updating frequency points of each iteration; performing second-stage iterative filtering according to the iterative record through an expansion filter; and updating the data after the second stage of iterative filtering, and outputting a noise substrate estimation result after the iteration is stopped. The invention can realize accurate approximation of the noise floor.
Description
Technical Field
The present invention belongs to the technical field of signal processing, and in particular, relates to a noise floor estimation method, apparatus, device, and storage medium.
Background
Noise floor refers to the power spectrum or background noise component in the spectrum of the acquisition signal. In wideband signal detection, it is often necessary to first estimate the noise floor and then set the size of the signal detection threshold based on the estimation result. The actual channel is often time-varying dispersive, and the background noise output by the receiver is not ideal Gaussian white noise due to the influence of impedance matching, nonlinearity of devices and power superposition of interference sources in the environment, so that an uneven noise substrate is generated, and the estimation of the uneven noise substrate is a key technology in broadband signal detection.
The common means in noise floor estimation is mainly nonlinear filtering based on morphological operation, and the morphological corrosion operation can weaken wave crests and strengthen wave troughs; the expansion operation can strengthen the wave crest and weaken the wave trough. The erosion and expansion operations are in fact equivalent to the sequencing filters, by cascading the sequencing filters with the erosion-expansion function, an open-close operation filter can be generated, the erosion filter being preceded by an open operation filter and vice versa. The noise floor estimation is generally implemented by an open operation filter, that is, a corrosion operation filter is used to filter out the signal wave crest, a noise floor is reserved, and then the expansion operation is used to smooth the noise floor estimation result.
Single morphological filtering based on fixed filter length generally makes it difficult to obtain accurate noise floor estimation, especially when the signal bandwidths within the system bandwidth differ significantly, too small filter length cannot filter out all signal interference, and too large filter length can severely degrade the resolution of the noise floor estimation. The existing solution is generally to stop iteration after finishing filtering all wave peaks by slowly increasing the filter length so as to keep better noise bottom resolution as much as possible.
However, the open operation filter is directly cascaded with corrosion and expansion operations with the same length, if the corrosion operation can not completely filter wave peaks, the expansion operation can re-strengthen and restore the wave peaks which are partially corroded, and only when the filter length is greater than the signal bandwidth, the signal wave peaks can be completely filtered, so that the noise bottom estimation method directly based on the open operation filter contains a large amount of redundant calculation, and is low in efficiency and difficult to guarantee in real-time performance. In addition, the traditional iterative convergence threshold calculation method based on the median or average value is poor in accuracy, is easy to be interfered by strong signals, and can influence the resolution of noise floor estimation.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a noise substrate estimation method, a device, equipment and a storage medium. And in the second stage, the iterative ordering filter with the expansion function is utilized to smooth the noise bottom, and the symmetry of the two-stage filter and the consistency of data updating are ensured through fine parameter setting, so that the accurate approximation of the noise bottom is realized.
The aim of the invention is achieved by the following technical scheme:
a method of noise floor estimation, the method comprising:
calculating a signal power spectrum in response to the acquired signal;
performing first-stage iterative filtering through a corrosion filter;
detecting spectral entropy white noise frequency points of signals;
calculating a data updating threshold and an iteration convergence threshold of the first stage of iterative filtering;
repeatedly executing the first stage iterative filtering, detecting spectral entropy white noise frequency points of signals, calculating a data updating threshold and an iterative convergence threshold of the first stage iterative filtering until corrosion filtering iteration is completed, and recording total iteration times and noise floor updating frequency points of each iteration;
performing second-stage iterative filtering according to the iterative record through an expansion filter;
and updating the data after the second stage of iterative filtering, and outputting a noise substrate estimation result after the iteration is stopped.
Further, the calculating the signal power spectrum specifically includes:
define 2N point intermediate frequency signal input as X= [ X ] 1 ,x 2 ,…x 2N ]Hamming window w= [ W ] defining the same length as 2N 1 ,w 2 ,…w 2N ]；
Windowing an input signal to obtain X w ＝[x 1 w 1 ,x 2 w 2 ,…x 2N w 2N ]；
For X w Fu Zuoli leaf transform then ignores the symmetrical negative frequency part, the resulting signal spectrum xf=fft (Xw) being a sequence of length N;
computing power spectrum estimate P 0 ＝|Xf| 2 。
Further, the performing the first stage of iterative filtering by the erosion filter specifically includes:
length m of corrosion filter for kth iteration k ＝m 1 ++ (k-1) delta, power obtained by iterating k-1Spectrum P k-1,1 Input corrosion filter R (n, m k ) Obtain output P k,1 ；
Wherein the initial length of the ordering filter is m 1 ，m 1 The step size for each iteration is an odd number, delta is an even number.
Further, the spectral entropy white noise audio point of the detection signal specifically includes:
generating a white noise sequence with the length of 2L, and calculating the power spectrum Z= [ Z ] of the white noise 1 ,z 2 ,…,z L ]The spectral entropy value h is defined as:
generating K groups of white noise and respectively calculating the spectral entropy values to obtain a spectral entropy sequence
Hw＝[hw 1 ,hw 2 ,…hw K ] T ；
The method for calculating the spectral entropy detection threshold comprises the following steps:
th_enp＝hw+3σ；
p obtained by current iteration k,1 For power spectrum P 0 Performing whitening treatment to obtain Pw k ：
Pw k ＝P 0 ./P k,1 ；
Calculating Pw by window length L k Obtaining a spectral entropy sequence H= [ H ] through sliding spectral entropy 1 ,h 2 ,…h N ] T When h i When the frequency point is more than or equal to th_ enp, the frequency point i is used as a white noise frequency point;
obtaining a white noise frequency point sequence number selected from the power spectrum in the kth iteration:
Y k ＝[y k,1 ,y k,2 ,…y k,M ]。
further, the calculating the data update threshold and the iteration convergence threshold of the first stage of iterative filtering specifically includes:
according to the chi-square distribution property, setting the confidence coefficient to be 0.9, and obtaining a convergence threshold th_stop of the kth iteration according to the sequence number of the white noise audio frequency point k ：
Setting the confidence coefficient to be 0.1 to obtain a kth data update threshold th_upd k :
Further, the performing the second stage of iterative filtering by the expansion filter according to the iterative record specifically includes:
the iterative filtering of the second stage is identical to the iterative filtering of the first stage in iteration times and the data updating is kept consistent, and the parameter of the iterative filtering of the kth stage is R (m k -n k +1,m k ) The input of the kth iteration of the second stage is P r,k-1 The expansion filter is R (m k -n k +1,m k ) The filtering output is P r,k 。
Further, the updating the data after the second stage of iterative filtering, and outputting the noise base estimation result after the iteration is stopped specifically includes:
the number set of the frequency points which are not updated in the kth iteration is I k ＝[i k,1 ,i k,2 ,…i k,D ]Frequency point of (1), I k ＝[i k,1 ,i k,2 ,…i k,D ]Resetting to the result of the corresponding frequency point in k-1 iterations:
P r,k (i k,j )＝P r,k-1 (i k,j ),j＝1,2,…,D；
after the data updating is completed, judging whether iteration is stopped, when k is<And (3) continuing the second stage of iterative filtering when r, stopping iteration when k=r, and outputting a noise floor estimation result P r,r 。
In another aspect, the present invention also provides a noise floor estimation apparatus, including:
the power spectrum calculation module is used for executing the steps of: calculating a signal power spectrum in response to the acquired signal;
the first stage filtering module is used for executing the steps of: performing first-stage iterative filtering through a corrosion filter;
the white noise frequency point detection module is used for executing the steps of: detecting spectral entropy white noise frequency points of signals;
the threshold calculating module is used for executing the steps of: calculating a data updating threshold and an iteration convergence threshold of the first stage of iterative filtering;
the first stage convergence judging module is used for executing the steps of: repeatedly executing the first stage iterative filtering, detecting spectral entropy white noise frequency points of signals, calculating a data updating threshold and an iterative convergence threshold of the first stage iterative filtering until corrosion filtering iteration is completed, and recording total iteration times and noise floor updating frequency points of each iteration;
the second stage filtering module is used for executing the steps of: performing second-stage iterative filtering according to the iterative record through an expansion filter;
the result output module is used for executing the steps of: and updating the data after the second stage of iterative filtering, and outputting a noise substrate estimation result after the iteration is stopped.
In another aspect, the present invention also provides a computer device, where the computer device includes a processor and a memory, where the memory stores a computer program, and the computer program is loaded and executed by the processor to implement any of the noise floor estimation methods described above.
In another aspect, the present invention also provides a computer readable storage medium having stored therein a computer program loaded and executed by a processor to implement any of the noise floor estimation methods described above.
The invention has the beneficial effects that:
(1) The invention can effectively reduce the iteration times of noise floor convergence and effectively improve the algorithm efficiency of static noise floor estimation on the premise of maintaining the performance.
(2) The invention can improve the accuracy of iteration threshold and the resolution of noise floor estimation.
(3) The invention can ensure the strict symmetry of the two-stage filter, the data updating and the iteration times are strictly consistent, so that the corrosion iteration-expansion iteration of the two stages has the same noise floor approximation capability as the one-stage open operation iteration filtering.
Drawings
FIG. 1 is a block diagram of a noise floor estimation method according to an embodiment of the present invention;
fig. 2 is a schematic flow chart of a noise floor estimation method according to an embodiment of the present invention;
fig. 3 is a block diagram of a noise floor estimation device according to an embodiment of the present invention.
Detailed Description
Other advantages and effects of the present invention will become apparent to those skilled in the art from the following disclosure, which describes the embodiments of the present invention with reference to specific examples. The invention may be practiced or carried out in other embodiments that depart from the specific details, and the details of the present description may be modified or varied from the spirit and scope of the present invention. It should be noted that the following embodiments and features in the embodiments may be combined with each other without conflict.
All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The open operation filter is directly cascaded with corrosion and expansion operations with the same length, if the corrosion operation can not completely filter wave peaks, the expansion operation can re-strengthen and restore the wave peaks which are partially corroded, and only when the filter length is larger than the signal bandwidth, the signal wave peaks can be completely filtered, so that the noise bottom estimation method directly based on the open operation filter contains a large amount of redundant calculation, and is low in efficiency and difficult to guarantee in real-time performance. In addition, the traditional iterative convergence threshold calculation method based on the median or average value is poor in accuracy, is easy to be interfered by strong signals, and can influence the resolution of noise floor estimation.
In order to solve the above technical problems, the following embodiments of the noise floor estimation method, apparatus, device and storage medium of the present invention are provided.
Example 1
Referring to fig. 1 and 2, fig. 1 is a flowchart of a noise floor estimation method according to the present embodiment, and fig. 2 is a flowchart of a noise floor estimation method according to the present embodiment. The method specifically comprises the following steps:
step S100: in response to the acquired signal, a signal power spectrum is calculated.
The step is realized by a power spectrum calculator, the power spectrum calculator carries out windowing short-time Fourier transform on an input signal with a certain point number and an overlapping rate, calculates to obtain the power spectrum of the broadband signal, and takes the power spectrum as the input of the subsequent noise floor estimation after linear smoothing processing.
Specifically, assume that the 2N-point intermediate frequency signal input is x= [ X 1 ,x 2 ,…x 2N ]Hamming window w= [ W ] defining the same length as 2N 1 ,w 2 ,…w 2N ]Windowing the input signal to obtain
X w ＝[x 1 w 1 ,x 2 w 2 ,…x 2N w 2N ]，
For X w Fu Zuoli leaf transform then ignores the symmetric negative frequency part, and the resulting signal spectrum xf=fft (Xw) is a sequence of length N, calculating the power spectrum estimate P 0 ＝|Xf| 2 Which is then taken as an input to an iteratively ordered filter.
Step S200: the first stage of iterative filtering is performed by an erosion filter.
The step is implemented by an erosion filter which orders the filtering R (n, m) to be equivalent to a one-dimensional morphological operation, and when n < (m-1)/2, the ordering filter has an erosion function.
In particular, unlike the conventional manner of cascading erosion and dilation filters, here iterative filtering is performedThe wave is divided into two stages, wherein the first stage only uses an erosion filter, so as to rapidly filter out all signal peaks. Assume that the initial length of the ordering filter is m 1 (odd), the increment step of each iteration is delta (even), then the erosion filter length m of the kth iteration k ＝m 1 A power spectrum P obtained by iteration k-1 is obtained by + (k-1) delta k-1,1 Input corrosion filter R (n, m k ) Obtain output P k,1 。
Step S300: detecting spectral entropy white noise frequency points of signals.
Specifically, the noise floor iteration threshold and the threshold for signal detection should both be calculated from the white noise power. Assuming that the signal in the system bandwidth is sparse, white noise audio points without signal occupation and interference in the power spectrum are detected based on sliding spectrum entropy, and then white noise power and data updating and iteration stopping thresholds are calculated based on the white noise audio points.
First, a threshold th_ enp for sliding spectrum entropy detection is calculated. the calculation of th_ enp is only needed to be carried out once, a white noise sequence with the length of 2L is generated by a random number generation method, and the power spectrum Z= [ Z ] of the white noise can be calculated by the same method as that in the step 1 1 ,z 2 ,…,z L ]The spectral entropy value h thereof can be defined as:
generating K groups of white noise by the same method and respectively calculating spectral entropy values to obtain a spectral entropy sequence Hw= [ Hw ] 1 ,hw 2 ,…hw K ] T The method for calculating the spectrum entropy detection threshold comprises the following steps:
th_enp＝hw+3σ；
wherein:
p obtained by current iteration k,1 For power spectrum P 0 Meter for whitening treatmentCalculating to obtain Pw k ：
Pw k ＝P 0 ./P k,1 ；
Calculating Pw by window length L k Obtaining a spectral entropy sequence H= [ H ] through sliding spectral entropy 1 ,h 2 ,…h N ] T When h i When the power spectrum is not less than th_ enp, the frequency point i is used as a white noise frequency point, and then the white noise frequency point sequence number selected from the power spectrum in the kth iteration can be obtained: y is Y k ＝[y k,1 ,y k,2 ,…y k,M ]。
Step S400: and calculating a data updating threshold and an iteration convergence threshold of the first stage of iterative filtering.
Specifically, the power of each frequency point in the Gaussian white noise power spectrum obeys chi-square distribution, the confidence coefficient is set to be 0.9 according to the chi-square distribution property, and the iteration convergence threshold th_stop of the kth iteration can be obtained based on the white noise audio point serial number obtained in the step 3:
correspondingly, the confidence coefficient is set to be 0.1, and the data update threshold th_upd of the kth iteration can be obtained k :
Step S500: repeatedly executing the first stage iterative filtering, detecting spectral entropy white noise frequency points of signals, calculating a data updating threshold and an iterative convergence threshold of the first stage iterative filtering until the corrosion filtering iteration is completed, and recording the total iteration times and the noise floor updating frequency points of each iteration.
Specifically, the k iteration corrosion filtering result is P k,1 With the previous iteration result P k-1,1 (i) Difference ΔP k The method comprises the following steps:
ΔP k ＝|P k,1 (i)-P k-1,1 (i)|
if the estimation result of the frequency point i and the frequency point of the kth-1 timeThe resulting difference of (a) is smaller than the data update threshold th _ upd k Resetting the bin i to the previous result:
recording the number of the frequency point which is not updated in the iteration as I k ＝[i k,1 ,i k,2 ,…i k,D ]Thus, the data update for the kth iteration is completed.
When max (DeltaP k )<th_stop k When the corrosion filtering iteration convergence is completed, stopping iteration, recording the total iteration times r=k, and entering step S600; otherwise when max (DeltaP k )≥th_stop k And when the etching filtering iteration process is continued, returning to the step S200.
Step S600: and performing second-stage iterative filtering according to the iterative record through an expansion filter.
The step is realized by an expansion filter, the expansion filter is used for sequencing and filtering R (m-n+1, m) and one-dimensional morphological operation are equivalent, when n < (m-1)/2, the sequencing filter has an expansion function, and the corrosion filter R (n, m) and the expansion filter R (m-n+1, m) are sequencing filters corresponding to each other.
Specifically, only the expansion filter is used in the second stage, in order to ensure accurate estimation of the noise floor, the iterative filtering of the first stage and the second stage needs to ensure a one-to-one correspondence, and the correspondence mainly comprises three aspects:
1. the iteration times are kept consistent, namely, the iteration times in the corrosion stage are r, the expansion stage does not need to judge by a specific threshold, and the corrosion stage can directly iterate r times.
2. The filter parameters remain corresponding, e.g. R (n) k ,m k ) Then the kth iteration in the expansion phase should use R (m k -n k +1,m k )。
3. The data updates remain consistent. For the same iteration order, which bins were updated by the first iteration stage, the second iteration stage should also update and update only those bins.
The first stage iterative filtering output is P r,1 . The input of the kth iteration of the second stage is P r,k-1 The expansion filter is R (m k -n k +1,m k ) The filtering output is P r,k 。
Step S700: and updating the data after the second stage of iterative filtering, and outputting a noise substrate estimation result after the iteration is stopped.
Specifically, the number set of frequency points not updated in the kth iteration of the corrosion stage is I k ＝[i k,1 ,i k,2 ,…i k,D ]The expansion phase also requires resetting these bins to the result of the corresponding bins in k-1 iterations:
P r,k (i k,j )＝P r,k-1 (i k,j ),j＝1,2,…,D；
after the data updating is completed, judging whether the iteration is stopped, and according to the first corresponding relationship in the step S600, when k<r, returning to the step S600 to continue the second stage iterative filtering; and when k=r, the iteration is stopped, and the noise floor estimation result P is output r,r 。
The sorting filter mentioned in this embodiment designates a sorting filter R (n, m), and the output of the filter is the nth value after the ascending sorting of the elements in the sliding window with length of m.
The iteration controller used in the implementation is mainly responsible for updating the data of the noise floor of each frequency point in the morphological iteration filtering process and controlling when to stop iteration. The core of the iteration controller is the calculation of an iteration threshold, and mainly comprises the calculation of a data updating threshold and an iteration stopping threshold. For the noise floor estimation of any frequency point, the noise floor estimation of the frequency point is updated only when the difference between the current iteration filtering result and the previous iteration filtering result of the frequency point is larger than a certain threshold th_upd, and iteration is stopped when the difference between the noise floor estimation of all frequency points in the power spectrum and the previous iteration filtering result is smaller than the threshold th_stop.
The noise substrate estimation method provided by the embodiment can effectively reduce the iteration times of noise substrate convergence, and effectively improve the algorithm efficiency of static noise substrate estimation on the premise of maintaining the performance. The method can improve the accuracy of the iteration threshold and the resolution of noise floor estimation. The method can ensure strict symmetry of the two-stage filter, and strict consistency of data updating and iteration times, so that the corrosion iteration-expansion iteration of the two stages has the same noise floor approximation capability as the one-stage open operation iteration filtering.
Example 2
Referring to fig. 3, shown in fig. 3 is a block diagram of a noise floor estimation device according to this embodiment, which specifically includes:
a power spectrum calculation module 10 for performing the steps of: calculating a signal power spectrum in response to the acquired signal;
the first stage filtering module 20 is configured to perform the steps of: performing first-stage iterative filtering through a corrosion filter;
a white noise point detection module 30 for performing the steps of: detecting spectral entropy white noise frequency points of signals;
a threshold calculation module 40, configured to perform the steps of: calculating a data updating threshold and an iteration convergence threshold of the first stage of iterative filtering;
the first stage convergence judging module 50 is configured to execute the steps of: repeatedly executing the first stage iterative filtering, detecting spectral entropy white noise frequency points of signals, calculating a data updating threshold and an iterative convergence threshold of the first stage iterative filtering until corrosion filtering iteration is completed, and recording total iteration times and noise floor updating frequency points of each iteration;
the second stage filtering module 60 is configured to perform the steps of: performing second-stage iterative filtering according to the iterative record through an expansion filter;
the result output module 70 is configured to perform the steps of: and updating the data after the second stage of iterative filtering, and outputting a noise substrate estimation result after the iteration is stopped.
The noise substrate estimation device provided by the embodiment can effectively reduce the iteration times of noise substrate convergence, and effectively improve the algorithm efficiency of static noise substrate estimation on the premise of maintaining the performance. The device can improve the accuracy of iteration threshold and the resolution of noise floor estimation. The device can ensure strict symmetry of the two-stage filter, and strict consistency of data updating and iteration times, so that the corrosion iteration-expansion iteration of the two stages has the same noise floor approximation capability as the one-stage open operation iteration filtering.
Example 3
The preferred embodiment provides a computer device, which can implement the steps in any embodiment of the noise floor estimation method provided in the embodiment of the present application, so that the beneficial effects of the noise floor estimation method provided in the embodiment of the present application can be achieved, and detailed descriptions of the foregoing embodiments are omitted herein.
Example 4
Those of ordinary skill in the art will appreciate that all or a portion of the steps of the various methods of the above embodiments may be performed by instructions, or by instructions controlling associated hardware, which may be stored in a computer-readable storage medium and loaded and executed by a processor. To this end, an embodiment of the present invention provides a storage medium having stored therein a plurality of instructions capable of being loaded by a processor to perform the steps of any one of the embodiments of the noise floor estimation method provided by the embodiment of the present invention.
Wherein the storage medium may include: read Only Memory (ROM), random access Memory (RAM, random Access Memory), magnetic or optical disk, and the like.
The instructions stored in the storage medium may perform the steps in any of the embodiments of the noise floor estimation method provided by the embodiments of the present invention, so that the beneficial effects that any of the embodiments of the noise floor estimation method provided by the embodiments of the present invention can be achieved, which are detailed in the previous embodiments and are not repeated herein.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, and alternatives falling within the spirit and principles of the invention.
Claims (10)
1. A method of noise floor estimation, the method comprising:
calculating a signal power spectrum in response to the acquired signal;
performing first-stage iterative filtering through a corrosion filter;
detecting spectral entropy white noise frequency points of signals;
calculating a data updating threshold and an iteration convergence threshold of the iterative filtering at the first stage according to the spectral entropy white noise frequency points;
repeatedly executing iterative filtering at a first stage, detecting spectral entropy white noise frequency points of signals, calculating a data updating threshold and an iterative convergence threshold of the iterative filtering at the first stage until the difference between the iterative result at any time and the previous iterative result is smaller than the iterative convergence threshold, completing the iterative filtering iteration, resetting the iterative result of any frequency point to be the previous iterative result when the difference between the iterative result of any frequency point in each iterative process and the previous iterative result is smaller than the data updating threshold, and recording the total iterative times and the noise bottom updating frequency point of each iteration;
performing second-stage iterative filtering according to the iterative record through the expansion filter, wherein the second-stage iterative filtering has the same iterative times as the first-stage iterative filtering, the data updating is kept consistent, and the parameters of the filter are kept corresponding;
and updating the data after the second stage of iterative filtering, and outputting a noise substrate estimation result after the iteration is stopped.
2. The noise floor estimation method of claim 1, wherein the calculating a signal power spectrum specifically comprises:
define 2N point intermediate frequency signal input as X= [ X ] 1 ,x 2 ,…x 2N ]Hamming window w= [ W ] defining the same length as 2N 1 ,w 2 ,…w 2N ]；
Windowing an input signal to obtain X w ＝[x 1 w 1 ,x 2 w 2 ,…x 2N w 2N ]；
For X w Fu Zuoli leaf transformation then ignores the symmetrical negative frequency part, resulting in a signalThe number spectrum xf=fft (Xw) is a sequence of length N;
computing power spectrum estimate P 0 ＝|Xf| 2 。
3. The noise floor estimation method of claim 2, wherein the performing the first stage iterative filtering by the erosion filter specifically comprises:
length m of corrosion filter for kth iteration k ＝m 1 A power spectrum P obtained by iteration k-1 is obtained by + (k-1) delta k-1,1 Input corrosion filter R (n, m k ) Obtain output P k,1 ；
Wherein the initial length of the corrosion filter is m 1 ，m 1 The step size for each iteration is an odd number, delta is an even number.
4. A noise floor estimation method according to claim 3, wherein the spectral entropy white noise audio points of the detection signal specifically comprise:
generating a white noise sequence with the length of 2L, and calculating the power spectrum Z= [ Z ] of the white noise 1 ,z 2 ,…,z L ]The spectral entropy value h is defined as:
generating K groups of white noise, and respectively calculating spectral entropy values to obtain a spectral entropy sequence:
Hw＝[hw 1 ,hw 2 ,…hw K ] T
the method for calculating the spectral entropy detection threshold comprises the following steps:
th_enp＝hw+3σ
p obtained by current iteration k,1 For power spectrum P 0 Performing whitening treatment to obtain Pw k ：
Pw k ＝P 0 ./P k,1
Calculating Pw by window length L k Obtaining a spectral entropy sequence H= [ H ] through sliding spectral entropy 1 ,h 2 ,…h N ] T When h i When the frequency point is more than or equal to th_ enp, the frequency point i is used as a white noise frequency point;
obtaining a white noise frequency point sequence number selected from the power spectrum in the kth iteration:
Y k ＝[y k,1 ,y k,2 ,…y k,M ]。
5. the method of estimating a noise floor according to claim 4, wherein calculating the data update threshold and the iteration convergence threshold of the first stage of iterative filtering specifically comprises:
according to the chi-square distribution property, setting the confidence coefficient to be 0.9, and obtaining an iteration convergence threshold th_stop of the kth iteration according to the sequence number of the white noise audio frequency point k ：
Setting the confidence coefficient to be 0.1 to obtain a data update threshold th_upd of the kth iteration k ：
6. The noise floor estimation method of claim 5, wherein said performing second stage iterative filtering by an expansion filter according to an iterative record specifically comprises:
the kth iteration filter parameter is R (m k -n k +1,m k ) The input of the kth iteration of the second stage is P r,k-1 The expansion filter is R (m k -n k +1,m k ) The filtering output is P r,k 。
7. The method for estimating a noise floor according to claim 6, wherein updating the data after the second stage of iterative filtering, and outputting the estimated result of the noise floor when the iteration is stopped, specifically comprises:
the number set of the frequency points which are not updated in the kth iteration is I k ＝[i k,1 ,i k,2 ,…i k,D ]Frequency point of (1), I k ＝[i k,1 ,i k,2 ,…i k,D ]Resetting to the result of the corresponding frequency point in k-1 iterations:
P r,k (i k,j )＝P r,k-1 (i k,j ),j＝1,2,…,D；
after the data updating is completed, judging whether iteration is stopped, when k is<And (3) continuing the second stage of iterative filtering when r, stopping iteration when k=r, and outputting a noise floor estimation result P r,r 。
8. A noise floor estimation device, the device comprising:
the power spectrum calculation module is used for executing the steps of: calculating a signal power spectrum in response to the acquired signal;
the first stage filtering module is used for executing the steps of: performing first-stage iterative filtering through a corrosion filter;
the white noise frequency point detection module is used for executing the steps of: detecting spectral entropy white noise frequency points of signals;
the threshold calculating module is used for executing the steps of: calculating a data updating threshold and an iteration convergence threshold of the iterative filtering at the first stage according to the spectral entropy white noise frequency points;
the first stage convergence judging module is used for executing the steps of: repeatedly executing iterative filtering at a first stage, detecting spectral entropy white noise frequency points of signals, calculating a data updating threshold and an iterative convergence threshold of the iterative filtering at the first stage until the difference between the iterative result at any time and the previous iterative result is smaller than the iterative convergence threshold, completing the iterative filtering iteration, resetting the iterative result of any frequency point to be the previous iterative result when the difference between the iterative result of any frequency point in each iterative process and the previous iterative result is smaller than the data updating threshold, and recording the total iterative times and the noise bottom updating frequency point of each iteration;
the second stage filtering module is used for executing the steps of: performing second-stage iterative filtering according to the iterative record through the expansion filter, wherein the second-stage iterative filtering has the same iterative times as the first-stage iterative filtering, the data updating is kept consistent, and the parameters of the filter are kept corresponding;
the result output module is used for executing the steps of: and updating the data after the second stage of iterative filtering, and outputting a noise substrate estimation result after the iteration is stopped.
9. A computer device, characterized in that it comprises a processor and a memory, in which a computer program is stored, which computer program is loaded and executed by the processor to implement the noise floor estimation method according to any of claims 1-7.
10. A computer readable storage medium, characterized in that the storage medium has stored therein a computer program, which is loaded and executed by a processor to implement the noise floor estimation method according to any of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210572981.1A CN114900246B (en) | 2022-05-25 | 2022-05-25 | Noise substrate estimation method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210572981.1A CN114900246B (en) | 2022-05-25 | 2022-05-25 | Noise substrate estimation method, device, equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114900246A CN114900246A (en) | 2022-08-12 |
| CN114900246B true CN114900246B (en) | 2023-06-13 |
Family
ID=82725757
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210572981.1A Active CN114900246B (en) | 2022-05-25 | 2022-05-25 | Noise substrate estimation method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114900246B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113670186A (en) * | 2021-07-16 | 2021-11-19 | 华东交通大学 | GNSS deformation monitoring denoising method combining variational modal decomposition and permutation entropy |
| CN114258567A (en) * | 2019-06-17 | 2022-03-29 | 弗劳恩霍夫应用研究促进协会 | Audio encoder, audio decoder and related methods and computer programs with signal dependent number and precision control |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| ES2422412T3 (en) * | 2008-07-11 | 2013-09-11 | Fraunhofer Ges Forschung | Audio encoder, procedure for audio coding and computer program |
| WO2018090467A1 (en) * | 2016-11-17 | 2018-05-24 | 丹阳华神电器有限公司 | Fuzzy entropy based noisy signal processing method and iterative singular spectrum soft thresholding denoising method |
| RU2744485C1 (en) * | 2017-10-27 | 2021-03-10 | Фраунхофер-Гезелльшафт Цур Фердерунг Дер Ангевандтен Форшунг Е.Ф. | Noise reduction in the decoder |
| CN109545188B (en) * | 2018-12-07 | 2021-07-09 | 深圳市友杰智新科技有限公司 | Real-time voice endpoint detection method and device |
| CN112737711B (en) * | 2020-12-24 | 2023-04-18 | 成都戎星科技有限公司 | Broadband carrier detection method based on adaptive noise floor estimation |
-
2022
- 2022-05-25 CN CN202210572981.1A patent/CN114900246B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114258567A (en) * | 2019-06-17 | 2022-03-29 | 弗劳恩霍夫应用研究促进协会 | Audio encoder, audio decoder and related methods and computer programs with signal dependent number and precision control |
| CN113670186A (en) * | 2021-07-16 | 2021-11-19 | 华东交通大学 | GNSS deformation monitoring denoising method combining variational modal decomposition and permutation entropy |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114900246A (en) | 2022-08-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111934711B (en) | Parameter estimation method of time-frequency aliasing frequency hopping signal | |
| CN103680515A (en) | Proportional adaptive filter coefficient vector updating method using coefficient reusing | |
| CN114900246B (en) | Noise substrate estimation method, device, equipment and storage medium | |
| CN109586728B (en) | Signal blind reconstruction method under modulation broadband converter framework based on sparse Bayes | |
| CN103701414B (en) | The selecting device of nonlinear terms and method, identification system and compensation system | |
| JP3918034B2 (en) | Method and apparatus for determining mask limits | |
| Bolic et al. | Performance and complexity analysis of adaptive particle filtering for tracking applications | |
| CN112396567B (en) | Scattered point cloud denoising method based on normal correction and position filtering two-step method | |
| CN112003588B (en) | Adaptive signal filtering method based on polymorphic variable step length normalized mean square | |
| US10393865B2 (en) | Phase retrieval algorithm for generation of constant time envelope with prescribed fourier transform magnitude signal | |
| CN112764345A (en) | Strong nonlinear system Kalman filter design method based on target state tracking | |
| CN102598126A (en) | Information processing device, auxiliary device therefor, information processing system, control method therefor, and control program | |
| CN110909635A (en) | Waveform similarity analysis method of nonlinear element model | |
| CN113299284B (en) | Speech recognition device, method, equipment and medium based on adaptive filtering | |
| CN108111448A (en) | Generation method, device and the pre-distortion calibration equipment of predistortion lookup table | |
| CN115079573A (en) | High-order expansion strong tracking filter of nonlinear system | |
| CN113452350B (en) | Variable step block sparse affine projection self-adaptive filter | |
| CN110457863B (en) | Parameter estimation method for wind driven generator pitch subsystem based on ellipsoid contraction filtering | |
| CN105448304A (en) | Voice signal noise spectrum estimation method and apparatus, and noise reduction processing method | |
| CN112255455A (en) | Signal processing method, signal processor, device and storage medium | |
| CN112929006A (en) | Variable step size selection updating kernel minimum mean square adaptive filter | |
| CN116127266B (en) | Secondary coefficient sparse method based on maximum inter-class variance | |
| CN112151053B (en) | Speech enhancement method, system, electronic device and storage medium | |
| CN110504948B (en) | Signal processing method and equal-interval fractional delay filter | |
| CN104242872A (en) | Signal processing method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |